Applying behavioural economics to public policy in Canada · Canadian public policy. We ﬁrst contextualize government policies that have incorporated insights from behavioural economics

Canadian Journal of Economics / Revue canadienne d’economique, Vol. 50, No. 3August 2017. Printed in Canada / Aout 2017. Imprime au Canada

0008–4085 / 17 / 599–635 / © Canadian Economics Association

Applying behavioural economics to publicpolicy in Canada

Robert French Harvard UniversityPhilip Oreopoulos University of Toronto; National Bureau of

Economic Research; Canadian Institute ForAdvanced Research

Abstract. Behavioural economics incorporates ideas from psychology, sociology andneuroscience to better predict how individuals make long-term decisions. Often the ideasadopted include present or inattention bias, both potentially leading to suboptimal out-comes. But these models also point to opportunities for effective, low-cost governmentpolicies that can have meaningful positive effects on people’s long-term well-being. Thelast decade has been marked by a growing interest from governments the world over inusing behavioural economics to inform policy decisions. This is true of Canada as well.In this paper we discuss the increasingly important role behavioural economics plays inCanadian public policy. We first contextualize government policies that have incorporatedinsights from behavioural economics by outlining a collection of models of intertemporalchoice. We then present examples of public policy initiatives that are based upon findingsin the field, placing particular emphasis on Canadian initiatives. We also document futureopportunities, challenges and limitations.

Resume. Appliquer l’economie comportementale a la politique publique au Canada. L’econo-mie comportementale incorpore des idees en provenance de la psychologie, la sociologie etla neuroscience afin de mieux predire comment les individus prennent des decisions a longterme. Souvent ces idees adoptees incluent le biais d’ancrage dans le present ou le biaisd’inattention – lesquels menent potentiellement a des resultats sous-optimaux. Mais cesmodeles soulignent aussi les possibilites de politiques gouvernementales effectives et a bascouts qui peuvent avoir des effets benefiques significatifs sur le bien-etre des gens a longterme. Il y a eu un interet croissant des gouvernements partout de par le monde au cours de

We are extremely grateful to Raj Chande (Behavioural Insights Team), Urvashi Dhawan-Biswal(Employment and Social Development Canada), Mireille Ethier (Canada Revenue Agency),Reuben Ford (Social Research and Demonstration Corporation), Elizabeth Hardy (PrivyCouncil Office), Julian House (Treasury Board Secretariat), Kory Kroft (University of Toronto),Kathryn Mills (Canada Revenue Agency), Hasti Rahbar (Employment and Social DevelopmentCanada), Aaron Rosenberg (Canada Revenue Agency), Steven Ryan (Abdul Latif JameelPoverty Action Lab), Dilip Soman (Behavioural Economics in Action at Rotman), Will Tucker(Ideas42), Sasha Tregebov (Deloitte) and Ekaterina Tyshchenko (Deloitte) for providing helpfulcomments and information. Opinions expressed in this paper are those of the authors only anddo not necessarily represent those of any government or institutional organization. Anyomissions or errors are, of course, the authors’ sole responsibility.Corresponding author: Philip Oreopoulos, [email protected]

600 R. French and P. Oreopoulos

la derniere decennie a utiliser l’economie comportementale pour informer les decisions depolitiques publiques. C’est vrai pour le Canada aussi. Dans ce memoire, on discute le rolede plus en plus important que joue l’economie comportementale dans la politique publiqueau Canada. D’abord on contextualise les politiques gouvernementales qui ont incorporedes intuitions en provenance de l’economie comportementale en declinant une collectionde modeles de choix intertemporels. Ensuite, on presente des exemples d’initiatives enpolitique publique qui sont fondees sur des resultats dans ce champ d’etudes, en mettantl’accent sur les initiatives canadiennes. On documente aussi les possibilites futures, les defiset les limitations de cette avenue.

1. Introduction

For decisions involving immediate costs and long-term benefits, standard in-vestment models in economics assume that individuals carefully consider everyexpected consequence from one action against every expected consequence fromeach alternative action. Take, for example, the choice that most commuters faceeach day on their way to work: the choice between walking up a set of stairs orstanding on an adjacent escalator. How does the commuter resolve this decision?Under the traditional rational investment model, the commuter evaluates theimmediate cost of taking the stairs against the long-term and uncertain healthbenefits of doing so (and likewise with the escalator). She then takes the actionwith the largest associated lifetime net benefit.

And yet, contrary to the traditional model’s predictions, people often responddramatically to small environmental changes that make one action more salientor attractive without significantly altering its consequences. In Hamburg, forinstance, commuters suddenly started taking the stairs to work after city officialspainted them in the design of a running track, complete with lane markers andpictures of other runners bounding forward.1 In Sweden, designing a set of stairsin the form of piano keys—that actually played musical notes—caused 66% morepeople to choose the stairs.2 In instances such as these, where the chief costs andbenefits have not changed, we often see a markedly different behaviour resultfrom supposedly irrelevant environmental changes, and perhaps for a very simplereason: it is fun to pretend you are racing along a track or playing the piano.3

While traditional economic models may have a difficult time explaining suchinterventions’ effects, the emerging field of behavioural economics seeks to re-fine the standard neoclassical assumptions of human behaviour in an attemptto better predict these apparent anomalous actions. To do this, behavioural eco-nomics incorporates findings from psychology, neuroscience and sociology into1 In 2015, a red running track was painted on top of a set of stairs in a new subway station,

Jungfernstieg, in Hamburg, Germany, with the goal of encouraging public transit use and theuse of the stairs.

2 In 2009, a small Swedish initiative sought to make the daily activity of taking the stairs more fun.A video of this particular initiative is available at youtube.com/watch?v=2lXh2n0aPyw.

3 Richard Thaler, in his book Misbehaving, terms such irrelevant environmental factors as SIFs,standing for “supposedly irrelevant factors.”

Behavioural economics in Canada 601

its models of human behaviour. These models often allow for deviations from thestandard assumption of economic rationality, capturing the cognitive and per-ceptual biases of our brains. Yet, at the same time, behavioural economics doesnot seek to reject neoclassical utility maximization models, but rather to refinethem, to incorporate human tendencies that are not always in our best long-runinterests, but nevertheless prevalent.

Incorporating more realistic assumptions of human behaviour into economicmodels can lead to profound differences in predicted actions, many of which areof interest to practitioners and policy makers. To this end, the field has attractedwide and growing attention: compared to traditional programs with the samegoals, interventions that draw from insights in behavioural economics may bemore cost effective, given that research suggests that even small changes in theway choices are presented or in the way information is conveyed can lead to largechanges in behaviour.4

As a consequence of this research, the last decade has been marked by a grow-ing interest from governments across the world in using behavioural economics toinform policy decisions. The pioneering example of government engagement withthis field of research is the UK’s Behavioural Insights Team (BIT). This team,known as the “nudge unit,” was established in 2010 after Richard Thaler andCass Sunstein’s book Nudge made it to the British Conservative Party’s summerreading list (Thaler 2015). The stated objective of the UK’s BIT is to spread theunderstanding of behavioural economics among British policy makers as well asto conduct trials and policy work that utilize findings from the field (Service et al.2014). After the first two years of the BIT’s inception, the team coined four prin-ciples that undergird their interaction with public policy and in turn succinctlysummarize the way in which policy makers the world over seek to incorporatefindings from behavioural economics into their policies more generally: (1) makeit easy, (2) make it attractive, (3) make it social and (4) make it timely (Serviceet al. 2014).

The appeal of applying behavioural economics to public policy has beenthe low costs associated with many of the behavioural economic interventions,the effectiveness of such interventions and the ease of testing these interven-tions through randomized control trials. That the UK’s BIT successfully intro-duced many cost saving and effective policies has served to propel the use ofbehaviourally based initiatives across a wide range of government departmentsin many countries; many government initiatives have explicitly stated that theyintend to model their respective behavioural insight teams after the UK’s BIT.5

Departments utilizing behavioural insights have subsequently developed in manycountries within both the private sector and the public sector, and with collabo-ration between the two sectors (Whitehead et al. 2014).

4 For a summary of such research, see Madrian (2014).5 See, for example, the official announcement concerning Ontario’s Behavioural Insights Unit

(Ministry of Finance 2015) and the White House’s Social and Behavioral Sciences Team 2015Annual Report (Social and Behavioral Sciences Team 2015).


Within the last few years behavioural economics has played an increasinglyimportant role in policy making within Canada. At the federal level, the Inno-vation Hub at the Privy Council Office was established in February of 2015. TheInnovation Hub is a resource that works alongside other federal departments, act-ing as a source of expertise on areas such as behavioural economics and designthinking. Similarly, the Canada Revenue Agency and the Ministry of Employ-ment and Social Development Canada have recently established Innovation Labs.The Canada Revenue Agency has been using insights from the behavioural eco-nomics literature for some time now and its Lab helps frame the experiments anddisseminate the results across the organization. Additionally, at the provinciallevel, Ontario’s Behavioural Insights Unit was officially established in 2015 andhas since conducted several successful pilot projects all based upon some of thecore principles of behavioural economics. We believe there is presently a growingenthusiasm for incorporating behavioural economics into public policy withinCanada.

The purpose of this paper is not to provide an exhaustive literature reviewof studies documenting behavioural economic insights, but rather to providean up-to-date report on the role that behavioural economics plays in Cana-dian public policy. In the next section, we contextualize government policiesthat have incorporated insights from behavioural economics by presenting arecent collection of behavioural economic models of intertemporal choice. Wethen provide a set of examples of policy initiatives that have incorporated in-sights from behavioural economics, with a particular focus on initiatives oc-curring in Canada. The set of examples is not intended to be comprehensive,but rather representative of the current policies that are being enacted withinCanada. Finally, we consider the Canadian experience of behavioural economicsin an international framework, comparing its experience to that of other coun-tries. This final section will also document future opportunities, challenges andlimitations to incorporating behavioural economics into public policy moregenerally.

2. Behavioural models of intertemporal choice

In this section, we outline three models that are frequently used in behaviouraleconomics as alternatives to the standard time-consistent intertemporal model ofutility maximization. We do this to contextualize the government policies we dis-cuss in the next section. All three models—present bias, inattentiveness and socialidentity, which incorporate tendencies observed in psychology and sociology—suggest new opportunities for effective, low-cost government policies that canhave meaningful positive effects on people’s long-term well-being. We also un-derscore how behavioural economics should not be perceived as a rejection ofneoclassical economics, but rather as an add-on that refines some of its underlyingassumptions about human behaviour.


Many decisions in life involve immediate costs and long-term, uncertain ben-efits. Many of them matter substantially over the long run, such as decisionsto smoke, exercise, use sunscreen, save, study and practice. Paul Samuelson, inhis classic 1937 article titled “A Note on Measurement of Utility,” proposedan approach for formalizing an individual’s decision-making process when con-sidering such intertemporal trade-offs. Samuelson, in formulating his model ofintertemporal choice wanted to extend Irving Fisher’s two period indifference-curve depiction of intertemporal choice into multiple time periods and, further,to show that “representing intertemporal trade-offs [in utility] required a car-dinal measure of utility” (Frederick et al. 2002, p. 355). Yet, despite his ferventreservations regarding the applicability of his model to questions of public policy,Samuelson’s formulation of intertemporal decision making became the bedrockof contemporary intertemporal utility maximization.6 The model’s parsimonyand convenience for working with dynamic optimization problems explains itsoverwhelming use within the economics profession (Frederick et al. 2002).

Consider a discrete-time version of Samuelson’s model, wherein each timeperiod the individual faces a binary choice:

V (x)=∑Tt=0±tu(x, t), where x ∈{0, 1} and 0 < ±�1. (1)

The crux of this model lies in ±t, the geometric discount rate. This captures allthe relevant information pertaining to the intertemporal portion of the individ-ual’s choice set. It is constant, implying that the individual compares taking thestairs now versus next week in the same manner she compares taking them a yearfrom now versus a year and one week. Importantly, decisions are time consistenton behalf of the individual; evaluated over any point in time, the individual isalways satisfied with her earlier decision

In discussing possible errors in intertemporal decision making, it helps to thinknormatively about how a person should trade off present versus future consump-tion to maximize her true well-being. Economists often assume individuals arebest off in this model with 0 < ±< 1, even though this implies that teenagers valuethemselves now more than themselves 20 years from now, and parents value theiryounger children more than their older children. While more challenging to workwith in solving dynamic optimization problems, O’Donoghue and Rabin (2001)argue that ± should be one. A ± value of one “is more in accord with the intu-ition that everybody...views as a more sensible welfare criterion: that we shouldwish on ourselves, our children, our neighbors and society the equal weighting ofthe expected hedonic well-being at different moments” (O’Donoghue and Rabin2001, p. 37). For the remainder of the section, assume that an individual indeedmaximizes lifetime welfare when ±= 1, though the discussion would remain thesame if choosing some other constant.

6 Samuelson rejected the idea that the discounted model of intertemporal utility maximization beused in policymaking. He famously concluded his 1937 article by noting that the idea of usingthe geometric discount rate to influence “ethical judgements of policy is one which deserves theimpatience of modern economists.”


To further simplify the exposition, and to bring the model to life, we also framethe decision regarding taking the stairs (x=1) versus the escalator (x=0) on ourdaily commute in terms of the model. Taking the stairs is costly; in the presentmoment we would much rather relax on an adjacent escalator. Yet, we know thatour future health, and thus well-being, will likely be higher if we do undergothe small cost of taking the stairs on a regular basis. Therefore, each time facingthis decision the individual must evaluate the immediate cost of taking the stairsagainst the corresponding long-term and uncertain health benefits of doing so.

Present bias:One of the most documented deviations in human behaviour from the ratio-nal economic agent is that we are often impatient. When contemplating futureevents, we tend to place greater weight on earlier events as they get closer tothe present moment, violating the neo-classical assumption of time consistency(O’Donoghue and Rabin 1999). O’Donoghue and Rabin (1999) document andlabel this behavioural insight, “present bias.”

Consider again our commuter deciding whether or not to take the stairs eachday. If the commuter has present-bias preferences, we can imagine them thinkingto themselves on Monday: “Tomorrow I will start taking those stairs” and onTuesday: “Tomorrow I will start taking those stairs,” and so on. Each day thepresent-bias commuter may earnestly want to start taking the stairs the followingday, yet when tomorrow is realized—because they place a greater weight on thepresent moment—the cost of taking the stairs outweighs any expected long-termbenefits of future good health.

David Laibson’s (1997) model of hyperbolic discounting attempts to describethis phenomenon more formally:

U (x)=u(x, 0)+¯∑T

t=0u(x, t), where 0�¯ < 1. (2)

Here, the ¯ captures the reduced weight placed on all future outcomes relativeto immediate ones, and a smaller ¯ implies higher impatience on behalf of thedecision maker. Deciding whether or not to take the stairs is a particularly inter-esting example to use when contemplating this model since the action of takingthe stairs only once is unlikely to make a long-term difference; the commuter hasto resolve to take the stairs not only that day but also virtually every day on theircommute in order for them to achieve future health benefits. We can thereforeempathize with the commuter who in the present moment neglects the option totake the stairs today in favour of tomorrow, especially when feeling hurried ortired.

This model highlights the difficulty traditional neo-classical models of in-tertemporal choice have in formalizing how large short-term psychic costs (asrepresented by ¯) of actions such as taking the stairs are compatible with individ-uals’ corresponding desires for future good health and well-being. If we assumeany realized costs of actually taking the stairs are small relative to the resultinghealth benefits of doing so, then to accommodate tendencies to procrastinate we


have to place a very large weight on the immediate cost of the action relative tothe same cost in the future. To do this would necessitate violating the neo-classicalassumption of time consistency inherent in traditional models of intertemporalchoice, highlighting an important conceptual distinction between traditional andbehavioural models of intertemporal choice.

Interestingly, BIs such as present bias are increasingly being grounded inneuroscience. There exists ample neurological evidence that the human braintreats decisions involving immediate consequences differently to ones that donot. Notably, McClure et al. (2007) report students’ brain activity during anexperiment that tested how these students responded to decisions involving dif-ferent time horizons. In the experiment, students arrived to a lab thirsty. Thesestudents were then asked to choose between having sips of juice/water at one oftwo specified times in the future. For instance, one of the choices involved choos-ing between one sip of refreshing juice in 20 minutes and two in 25 minutes, andanother choice was between one sip of juice now and two sips of juice in fiveminutes.

Under the assumption of geometric (or no) discounting, these two choices areidentical: in both cases, waiting an additional five minutes provides an extra sipof juice. Yet, the results from these two particular choices are markedly different.When deciding between one sip in 20 minutes and two sips in 25 minutes, roughly70% of the students were willing to wait the extra five minutes for an additionalsip of juice. In contrast, when deciding between one sip of juice immediately andtwo sips of juice in just five minutes time, only around 40% of the students electedto wait the additional five minutes for an extra sip of juice.

Furthermore, in addition to waiting up to 25 minutes for a sip of juice orwater—and perhaps to the displeasure of the participants—the students wereasked to make their decisions while in an MRI scanner, allowing the researchersto examine which areas of the students’ brains were being stimulated during thedecision making processes. The MRI scans indicated that the brain activity ofthe students when making the decision involving the distant sips of juice wasstrongest in the prefrontal cortex (the area of the brain most associated withmore complex decision making), whereas brain activity of the students whenmaking the decision involving the opportunity to have an immediate sip of juicewas strongest in the amygdala (an area of the brain associated with emotionalresponses).

While economists are not necessarily interested in the underlying mechanismsbetween causal interactions, grounding economic decision making processes inneuroscience can help better model the economic relationships they are interestedin (Krajibich and Dean 2015). Furthermore, as in this example of present-biaspreferences, evidence from neuroscience can help ensure that one’s postulatesregarding human behaviour are applicable across a variety of settings; neuro-science can provide a solid foundation for further economic investigation byrefining our underlying assumptions about human behaviour.


Inattention:While impatience may be a product of our neurological networks processing de-cisions involving immediate outcomes differently to those involving future ones,other suboptimal economic behaviour may result from our brain simply not pay-ing attention. Sometimes we may overlook a decision because we are unaware ofit or we discern (consciously or subconsciously) the cost of deliberating upon adecision to be too high relative to any benefits associated with the decision itself.7

Suboptimal economic behaviour may thus arise when particular options are notsalient.

Xavier Gabaix (2014) incorporates this idea into the intertemporal model ofutility maximization in a tractable manner:

U (x, y)=∑Tt=0u(mx, y, t), where m∈ (0, 1). (3)

In this model, there are two possible actions: x and y. We assume that yis an outcome associated with an inherently salient decision, such as where togo after ascending either the stairs or the escalator. In contrast, assume that xis a decision that is unimportant relative to our final destination such as thecommuter’s decision itself. Specifically, the model places an attention weight, m,on the action x. In the extreme case when m= 0, we just do not think about theaction x as an option.8

This model can be related to the work of Daniel Kahneman, who, in his bookThinking Fast and Slow discusses how we tend to treat decisions of various com-plexities with two different cognitive “systems.” Building off the work of previouspsychologists, Kahneman (2011) notes that we tend to make simple decisions(such as whether to take the escalator) automatically and that these automaticdecisions are conducted by our “system 1,” without much deliberate thought.This is in contrast to our “system 2,” which is preserved for more complex de-cisions and operates under our direct attention. That the decisions made by our“System 1” are automatic, they tend to be determined by habits and heuristics.Thus, if our habits or heuristics do not align with our long-term best interests, wemay be unwittingly enacting suboptimal economic behaviour; if we were forcedto stop and consider the trade-off between taking the stairs or the escalator, wemay act differently than if we were acting under the guidance of our heuristics.

Additionally, recent research by Mullainathan and Shafir (2013) points to thefact that inattention to decisions involving trade-offs—including intertemporalones—is exacerbated when individuals have less cognitive “bandwidth” to devote

7 For an interesting attempt to incorporate these ideas into neo-classical theory, see (Woodford2012).

8 Notice here that the attention weight, m, is exogenous to the individual’s utility function inequation (3), even though we consider it to be something that is determined (consciously orsubconsciously) by the individual herself. With m being exogenous, it is thus possible for us totalk about the individual acting suboptimally with respect to her true lifetime utility function, incontrast to when m is endogenous and consequently set at an optimal level by the individual ineach time period.


to thinking about said decisions.9 They suggest that an individual’s bandwidth isparticularly affected by stress (e.g., from money, time or family), which in turncan be directly related to external factors such as poverty. Through this lens,stressful circumstances make it hard for us to contemplate decisions, forcing usto rely on heuristics and habits, which in turn may explain much non-rationaleconomic behaviour.

Social identity:Social psychologists often emphasize the enormous role social identity plays inshaping preferences. They argue that questions like “What kind of person amI?” and “What are others like me doing?” serve as powerful reference points fordeciding how to act. Akerlof and Kranton (2000) formalize this tendency in asimple theoretical model that incorporates into an individual’s utility functiona social identity function, in which deviations in one’s actions from their ownreference group’s actions creates dissonance and declining utility. Here we com-bine a simplified version of this social identity function with Laibson’s modelof present-bias preferences to highlight how immediate social influences cou-pled with an overemphasis of the present can have a very large impact on one’seconomic decisions. Formally, consider:

V (x)=u(x, t, Ä(x, xg)

)+¯∑T

t=1u(x, t, Ä

(x, xg

)), (4)

where we define Ä(x, xg) as the social identity function which is a decreasingfunction of |x− xg|, and xg is the mean action performed by the individual’s peergroup. Individuals receive utility from not only how their own actions impactthem directly but also how those actions relate to what others in a group theyidentify with typically take or consider desirable.

If a present-bias individual discounts the future and cares a lot about how shefits in socially, what others around her are doing or what advice she gets from hergroup could be extremely important to her actions now, while having long-termconsequences later. For example, surrounded by co-workers that celebrate takingthe stairs each day instead of the escalator can lead to long-term health benefitsthrough a desire to share common social experiences.

Implications:The behavioural economic models presented above, which all deviate from the onein which individuals maximize their true long-term welfare, suggest cost-effectiveopportunities for government policy to improve lives. They formally suggest manydifferent ways in which governments can affect change without altering the chiefcosts and benefits of individuals’ choice sets. They provide an economic lens tointerpret the effectiveness—or lack thereof—of such policy interventions, often

9 Mullainathan and Shafir (2013) use “bandwidth” as an umbrella term to measure our“computational capacity, our ability to pay attention, to make good decisions, to stick with ourplans, and to resist temptations.”


incorporating practices from marketing and other consumer behaviour drivenfields of inquiry into a formal economic framework.

In this regard, we consider two broad types of government interventions. Firstlet us consider choice limiting interventions. By way of example, if we assume thattaking the stairs as opposed to the escalator is in most people’s long-term bestinterest, then a choice limiting intervention would be to ban the use of the esca-lator. Aside from not respecting the individuals for whom the escalator is clearlythe better choice (e.g., person with a broken leg or who just finished a marathon),this intervention removes the choice of the escalator entirely. In addition to anymoral considerations, as economists we are reluctant to impose choice constraintson individuals when tradeoffs exist, especially when these tradeoffs are not fullyobserved by the policy maker. This does not mean that bans or mandates do notoccasionally serve a purpose in public policy (consider seat belt use mandates),but improving individuals’ behaviour through such means usually comes at theexpense of making other individuals worse off.

Consider instead choice-preserving interventions, where particular actionsby the individual are encouraged without removing the individual’s liberty tochoose alternative actions. Such choice-preserving policy interventions typicallyprompt people to think of the benefits of a particular action or make a particularaction appear more appealing than the alternatives without actually altering anyoutcomes of the options available to the decision maker. Cass Sunstein uses anapt example of a GPS device in one’s car to illustrate this idea. He notes that“a GPS steers people in a certain direction, but people are at liberty to selecttheir own route instead” (Sunstein 2014, p. 2). It is these choice-preserving policyinterventions that Richard Thaler and Cass Sunstein termed as “nudges.”

However, one criticism often raised against the use of nudging in public policyis the potential for manipulation since individuals may not be aware that they arebeing encouraged to act in a particular manner. That nudges have the potential toencourage behaviour that is in conflict with their own long-term interests furthervalidates this concern. Yet, a potential way of assuaging this criticism is to makethe use of the nudge public knowledge. For instance, Sunstein (2014) maintainsthat any official nudge should be transparent and open; if a nudge is openlysubject to public scrutiny, the chances of manipulation are minimized.

In contrast, an argument defending the use of nudging in public policy isthat organizations or policy makers cannot avoid nudging, regardless of theirpolicy proposal; every policy intervention is embedded in a framework that isnot directly considered part of the intervention itself. For example, when a newpolicy is introduced, it is often obliged to feature a default option. Behaviouraleconomics just takes this one step further and carefully considers what this defaultoption should be in order to maximize the welfare of the program’s participants,given knowledge of cognitive and perceptual biases present in human behaviour.

Finally and perhaps most important, the very fact that nudges do influencebehaviour may validate their use. It suggests that people do overreact to im-mediate circumstances and thus may deviate from the actions that are in accord


with the standard intertemporal investment model. Given this discussion though,one thing is certain: nudges cross many different boundaries such as economics,politics and ethics and must be considered carefully before being implemented.

3. Interventions

In this section, we present policies and pilot projects that are based, at least inpart, upon the models outlined in section 2. We place an emphasis on Canadianinterventions, and while our examples are not comprehensive, we believe theyare representative of the initiatives currently under way in Canada. We focus onthe following areas of public policy: saving, health, education, employment andtaxation.

Saving:One of the most well-known applications of behavioural economics concernssaving for retirement. When deciding how much to save, the standard economicinvestment model assumes that individuals are forward-looking, able to forecasttheir needs and face little difficulty following through with their plans. Severalstudies note, however, that the behaviour of at least some people deviate fromthis model (e.g., Benartzi and Thaler 2007). Many seem to recognize this andreport feeling they are not saving enough. For instance, Choi et al. (2002) presentresults from a survey at a large US food corporation showing that, out of every100 respondents, 68 believe their savings rate to be too low. Yet, of these 68individuals, only 24 plan to increase their savings rate in the coming months, andfurther still, only three of these 24 respondents actually do.

A very successful approach used to encourage more savings among workers isto change the default enrollment option in employer pension programs.10 Whenemployees have to opt into their employer’s pension plan, take-up tends to bemuch lower than when employees are enrolled automatically. This is especiallytrue within the earlier years of employees’ tenure at a firm.11 Madrian and Shea(2001) document this phenomenon for a large US company that introduced au-tomatic enrollment into its 401(k) plan in 1998 (with the option to opt out). Theauthors found that participation in the company’s plan was 38% higher amongemployees who were automatically enrolled. Further, after accounting for thedifference in tenure and demographic characteristics between the group prior toautomatic enrollment, who could not join until one year of tenure at the firm, andthe group after, who were automatically enrolled when they started working at the

10 One cannot rule out the possibility of changes to other savings accounts negating any increase insavings induced by workplace pension enrollment. However, recent evidence from Canadasuggests crowd-out rates from workplace pension plans may be significantly less than one onaverage; Messacar (2015) estimates that a $1 increase in employer contributions to workplacepension plans crowd out private savings by $0.5 on average.

11 For a summary of the empirical literature surrounding this phenomenon, see Beshears et al.(2009).


2040

6080

100

Per

cent

age

of p

artic

ipat

ing

elig

ible

wor

kers

2006 2008 2010 2012 2014

Public sector Private sectorAll

FIGURE 1 Automatic enrollment and workplace pensions in Great Britain for eligible employeesparticipating in workplace pensions, by sectorNOTES: Vertical line indicates initiation of mandated automatic enrollment into workplacepensions. This graph is a replication of figure 5 in National Audit Office (2015).SOURCE: Department for Work and Pensions estimates derived from the Office for NationalStatistics’ Annual Survey of Hours and Earnings, Great Britain, 2004–2014

firm, the estimated comparable gap in participation rates increased to a remark-able 50%, showing the power that default enrollment options have on saving rates.

The results from this study are indicative of other, larger initiatives in retire-ment savings plans the world over. Perhaps most notable is the UK’s 2008 PensionAct. Concerned by declining saving rates among working adults, the UK phasedin a requirement whereby employers are mandated to enroll their employees into aworkplace pension plan (National Audit Office 2015).12 After becoming enrolled,employees can opt out of their workplace pension plan, yet so far only between 8%and 14% have chosen to do so, leading to a huge increase in the number of privatesector workers being enrolled in workplace pensions (figure 1). The legislation isestimated to have encouraged an additional £6.6 billion into workplace pensionssince its first two years of implementation, and this number is predicted to increaseto between £14 billion and £16 billion by 2019–2020 (National Audit Office 2015).

Motivated by similar concerns of undersaving, the Canadian government an-nounced in June its plan to enhance the Canada Pension Plan (CPP) (Department

12 The minimum contribution rate for employers is currently 1% of each employee’s salary but isincreasing to 2% in October 2017 and 3% in October 2018 (National Audit Office 2015).


of Finance Canada 2016).13 The federal government is seeking to increase theCPP’s replacement rate from 25 to 33.3% and to expand the maximum amountof income subject to CPP contributions, from $54,900 to $82,700 (Departmentof Finance Canada 2016, Milligan 2016). Support for the CPP enhancement hasbeen buttressed by research suggesting that increases in workplace pension plansin Canada do not fully crowd out private savings and that such increases are moreeffective in raising saving rates among individuals with weaker saving histories(Messacar 2015).

Another means to increase saving among working Canadians is to encouragethe use of private workplace pension plans, whereby employers and employeescontribute to a company fund.14 An immediate response to this goal is to man-date automatic enrollment into workplace plans, as has occurred in the UK.However, in Canada, there are currently legislative barriers to imposing an opt-out default policy for workplace pension plans (Palameta et al. 2011). One alter-native approach to encourage participation is to ensure every employee makesan affirmative choice between whether or not to enroll in any such program. Notchoosing is not an option. This approach necessarily makes the decision moresalient, discouraging the use of heuristics and combatting tendencies to procrasti-nate. Such choices have been termed “active choices” or “active decisions” withinthe behavioural economics literature.

Carrol et al. (2009) demonstrate the effectiveness of “active choice” in pro-moting participation in workplace pension plans. The authors analyzed a largefinancial services firm that changed its 401(k) enrollment policy. Prior to Novem-ber 1997, new hires to this firm were required to submit a form indicating their401(k) enrollment preference. In contrast, after November 1997, new hires didnot receive this form but were instead given a toll-free phone number to call if theywanted to enroll in the firm’s plan. The former enrollment process required em-ployees to make an active choice between either being enrolled in the plan or not,whereas the latter enrollment process assumed a default of non-enrollment. Theauthors show that after three months of employment, participation rates were 28percentage points higher among employees hired prior to November 1997 (thosewho had to make an active choice) and that this gap persisted between for at least42 months after being hired.

The goal of Canadian policy makers to increase saving rates is not limited toretirement savings either. The Canadian government is actively encouraging fam-ilies to save more for their children’s future education. As part of the RegisteredEducation Savings Plan (RESP), a tax-deferred savings account for postsec-ondary education, the federal government offers low-income families up to $2,000in the form of a grant, called the Canada Learning Bond (CLB). However, despite

13 Many studies have analyzed undersaving in Canada. See, for example, Wolfson (2011) andMoore et al. (2010).

14 Prior to the announcement by the federal government that they are to work on enhancing theCPP, the Ontario government sought to mandate enrollment into workplace pension plans,through the Ontario Retirement Pension Plan (ORPP).


FIGURE 2 SmartSaver’s online RESP enrollment procedureSOURCE: smartsaver.org (accessed July 2017)

no minimum contribution needed to receive the grant, it is estimated that in 2008only 16% of eligible households had claimed their CLB (Nayar Consulting 2013).

A lack of awareness of the RESP as well as the complexity of its signup pro-cess contribute to these low participation rates (Nayar Consulting 2013, Milligan2002).15 For example, parents are required to apply for and provide a socialinsurance number for their children and meet with a private bank representativeto process a lengthy application. Here, seemingly innocuous up-front costs maybe unintentionally affecting individuals’ propensity to take actions that are inaccord with their long-term best interest. Working with these setup constraints,the Omega Foundation established a national program called SmartSAVER thatencourages eligible families to claim the CLB by simplifying the application pro-cess and placing more onus on banks to complete the eligibility process. To dothis, SmartSAVER first collaborates with both private and public organizations topromote awareness of the CLB and RESP through numerous marketing schemes.

15 Another potential contributor to these low take-up rates is that when the CLB was introduced in2004, only children born after the introduction of the policy were eligible for the grant. Sincemany parents may consider saving for their children’s education more seriously when theirchildren are above the age of four, this feature of the program may have inadvertentlycontributed to the low take-up rates recorded in 2008.


Second, it simplifies the RESP application process by taking applicants throughan online application with five simple steps (figure 2). Third, upon completingthe online application, an experienced bank representative contacts the potentialclient to arrange an appointment and finalize registration, further simplifying theenrollment process by turning the onus upon the financial institution to completethe application (Kenter 2015).

Throughout the pilot project in Toronto, CLB take-up rates increased from27.7% in 2009 to 39.3% in 2012 (Nayar Consulting 2013). Yet, there remains a widedisparity in CLB take-up rates across regions in Canada, signalling potential forimprovement; take-up rates are as low as 1% in areas of Manitoba and as high as48% in regions of Toronto, where the initiative first began (SmartSAVER 2014).Government “Nudge Units” such as Ontario’s Behavioural Insights Unit andthe Privy Council’s Innovation Hub are currently exploring ways to simplify theCLB take-up process, which may require coordination with the Canada RevenueAgency to help verify low-income status. A system by which Canadian newbornand landed immigrant children are automatically enrolled in an RESP, withregular updates sent to parents and easy ways to deposit, would circumvent manyissues related to the program’s salience and enrollment process. In the meantime,the Omega Foundation’s SmartSAVER program provides a good example forhow an organization can help encourage greater take-up of public services whileworking with existing legal and procedural constraints.

Health:There is a long waiting list for organ donors in many countries, including Canada(Shimazono 2007, Canadian Institute for Health Information 2016). In 2014, forexample, 278 Canadians died while waiting for an organ transplant (CanadianInstitute for Health Information 2016). Survey evidence from Ontario suggests amajority of citizens are willing to register as donors, but only 29% of those eligibleare actually registered (Trillium Gift of Life Network 2016). Our models frombehavioural economics suggest that the upfront discomfort from thinking aboutdonating (in case of death) may prevent lives from being saved or that the act ofdonating simply is not a salient option for consideration. One policy tool oftenused to combat these potentialities is “presumed consent.” Individuals residing incountries with a policy of presumed consent are automatically registered as organdonors and have to purposefully opt out of this default position if they decideagainst being a donor. Many studies including Johnson and Goldstein (2003)have shown that organ donation consent rates are much higher in countries withpolicies of presumed consent than those without (figure 3).

Efforts to increase organ donation consent rates have also occurred in Canadawithin the province of Ontario. Yet, though a policy of presumed consent islikely to increase consent rates, a survey conducted by the market research firmIpsos shows that the large majority of Ontarians do not favour such a system;Ontarians appear to view organ donation as a personal choice and prefer thecurrent opt-in default (Trillium Gift of Life Network 2015). In fact, Ontario’s opt-


0 20 40 60 80 100Consent rate

Pre

sum

ed c

onse

ntE

xplic

it co

nsen

t

Sweden

Portugal

Poland

Hungary

France

Belgium

Austria

UK

Netherlands

Germany

Denmark

FIGURE 3 Organ donation consent rates, by countrySOURCE: Replication of figure 2 from Johnson and Goldstein (2003)

in default policy is situated within a prompted-choice system: when conducting ahealth card, driver’s licence or photo card transaction at a ServiceOntario centre,Ontarians are asked by a customer service representative whether they would liketo register as a donor.16

Conscious of the possibility of behavioural barriers affecting donation rates,Ontario’s recently established Behavioural Insights Unit, in collaboration withthe Behavioural Economics in Action at Rotman Centre (BEAR), partnered withthe Trillium Gift of Life Network, Ontario’s Ministry of Health and Long TermCare and the Ministry of Government and Consumer Services to test variousinterventions aimed at increasing consent rates within the confinements of theprovince’s prompted-choice system. The interventions focused on increasing thesalience of the long-term benefits, the simplicity of the registration process by al-tering the registration form itself (figure 4) and the time at which potential donorsreceived the registration form during their visit to a ServiceOntario centre.17 Todetermine the most effective registration process, these interventions were trialedacross six different treatment periods at a single ServiceOntario centre and the

16 Ontarians can also register to be a donor online, but only 15% of registrants do so (TreasuryBoard Secretariat 2016).

17 There were four such variations of the registration form, each including a different nudgestatement. One of the treatments also included a brochure including information pertaining toorgan donations.


(a)

(b)

FIGURE 4 Change in Ontario’s organ donation consent formNOTES: 4(a): Ontario’s original organ donor consent registration form. 4(b): An experimentalorgan donor consent registration form used in a pilot project conducted by the Ontario Ministry ofHealth, Trillium Gift of Life Network, Service Ontario and the Treasury Board Secretariat’sBehavioural Insights Unit. This experimental consent form (condition 4) simplifies the originalform and includes the nudge statement “If you needed a transplant, would you have one?”SOURCE: Treasury Board Secretariat (2016)


results from the treatment periods were compared to a control period at that sameServiceOntario centre (Treasury Board Secretariat 2016, Robitaille et al. 2016).

On average, organ donation rates for individuals assigned to any one of thetreatments increased by up to 143%. The most persuasive nudge statements askedpotential donors to imagine herself or a loved one needing an organ donation.And as a consequence of these results, the most effective organ donation regis-tration form has been adopted across all ServiceOntario Centres as of June 2016(Treasury Board Secretariat 2016, Robitaille et al. 2016).

Another public health issue that has been of interest to governments in Canadais influenza vaccination rates. This is because each year an average of 3,500 Cana-dian residents die from influenza (Gionet 2015). For most provinces and terri-tories in Canada, influenza vaccination rates increased between 2003 and 2013–2014 (Gionet 2015). However, in Ontario the opposite is true: the vaccination ratedecreased from 38% in 2003 to 34% in 2013–2014 (Gionet 2015). One factor thatmay be in part responsible for the lower vaccination rates in Ontario over thisperiod is that unlike most other provinces and territories in Canada, until late2012, it was illegal for pharmacists to administer the influenza vaccine (CanadianPharmacists Association 2012). By introducing legislation in 2012 that allowedpharmacists in Ontario to administer the influenza vaccine, the Ontario govern-ment made it more convenient for individuals to receive the vaccine, reducing theup-front costs for the recipients.

The change in Ontario’s legislation is in accord with other efforts and researchthe world over to make flu shots more convenient and more salient.18 Notably,Milkman et al. (2011) report the results of a field experiment at a large US firmthat tested to see how prompting employees to write down the date and/or timeof their respective free on-site influenza clinic affected vaccination rates. For thisfield experiment, all employees at the firm received an email informing them oftheir respective on-site vaccination clinic. In the email, some employees were alsoprompted either to write down the date or to write down the date and time of theirvaccination clinic. Vaccination rates between these three groups differed mean-ingfully, with those receiving the prompt to write down both the date and timeof their respective clinic having a 4.2 percentage-point higher vaccination rate.

Education:A discipline that has received less attention from behavioural economists is edu-cation, despite the fact that youth, with their developing brains, are particularlypredisposed to inattention and present bias (Lavecchia et al. 2016). The researcharea is growing, however, and there have been increasing attempts to nudge chil-dren and their parents toward decisions that promote education attainment andacademic performance.

18 See Chen and Stevens (2016) for a discussion on using behavioural economics to increasevaccination rates, as well as Corace et al. (2016) for a review of different methods used toincrease vaccination rates among healthcare workers.


We first examine an intervention that simplified the submission of the USFree Application for Federal Student Aid (FAFSA) form. All college and uni-versity students in the United States must complete the FAFSA in order to re-ceive public financial assistance. The application process itself, however, is notstraightforward, especially for students left on their own to navigate through it.Students must learn how to access the form and collect the information needed tocomplete it, including information about parents’ social security numbers, exactincome reported on their most recent tax files, vehicle ownership and other assets.Concerns regarding the complexity and inconvenience of filing the FAFSA haveprompted calls to amend the process (ACSFA 2005, Dynarski and Scott-Clayton2006). To explore how much difference application simplification could make,Bettinger et al. (2012) conducted a field experiment in conjunction with H&RBlock, known as the FAFSA study.

The FAFSA study identified eligible parents visiting H&R Block with chil-dren in their senior year at high school.19 After agreeing to participate in theexperiment, parents were designated into one of two treatment groups and a con-trol group. Individuals in the first treatment group (the “information treatment”)were provided with an estimate of the amount of financial aid their childrencould obtain from completing the FAFSA, along with information about tuitioncosts at four nearby colleges. Parents in the second treatment group (the “per-sonal assistance treatment”) were provided the same information as those in theinformation treatment and also offered direct assistance with completing the ap-plication. Since many of the questions to complete the FAFSA form are the sameas those asked on the tax form, the assistance took roughly eight extra minutesin the H&R Block office.

As can be seen in figure 4, personal assistance in the FAFSA applicationprocess had a considerable impact on both the FAFSA submission rates andthe propensity for individuals to attend a postsecondary institution. High schoolseniors with parents receiving the FAFSA application assistance had FAFSAfiling rates 16 percentage points higher than the control group (56% versus 40%).The assistance would have increased rates even more if the FAFSA did not requirethe students’ signature. Instead, H&R Block sent the complete, or near completeFAFSA home first, with a prepaid envelope and instructions for the student tosign and mail to the Department of Education. Still, when looking at enrollmenteffects, those from the assistance group were 8 percentage points more likely toattend a postsecondary institution than those from the control. The study alsofound similar affects for attending at least two years of college, even thoughthe treatment helped with entry for only the first year. Interestingly receivinginformation pertaining to the amount of financial aid one is likely to receive as wellas the cost of nearby colleges, had no significant effect on FAFSA applicationsor postsecondary enrollment. Information on its own was not enough to change

19 The study also found college enrollment effects for a separate sample of individuals visitingH&R Block with not more than a high school education and who were offered FAFSAcompletion assistance.


0.2

.4.6

Pro

babi

lity

filed

FA

FSA

0.2

.4.6

Pro

babi

lity

enro

lled

in c

olle

ge/u

nive

rsity

Control Personal assistance Information

FIGURE 5 Outcomes during first year following experimentNOTES: These bar charts display results of the FFSA experiment as reported in table 3 of Bettingeret al. (2012). Treatment effects are estimated using OLS. The Personal assistance treatment groupwas offered direct assistance with completing a FAFSA application. The Information treatmentgroup was provided with an estimate of the amount of financial aid their children could obtain fromcompleting the FAFSA. Participants (N D 868).SOURCE: Bettinger et al. (2012)

behaviour. The study demonstrates a nudge with personal interaction may bemore effective (yet more costly)—a result we return to below.

Applying for financial aid is just one step of many that must be taken to attendcollege or university. When transitioning to postsecondary education, studentsmust also determine which schools and programs to apply to, pay applicationfees, make additional financial arrangements and modify their daily routine—allof which can pose barriers to those not familiar with the process. To help over-come these, Oreopoulos and Ford (2016) developed a program called LifeAfter-HighSchool (lifeafterhighschool.ca) that integrates both program and financialaid application assistance into the grade 12 curriculum at low-transition highschools in Ontario.20 The program provides all high school seniors with in-classassistance, regardless of their current postsecondary plans. The slogan of theprogram is Keep your options open. Even for students not intending to continuetheir education, an offer of acceptance from a program that students themselves

20 Low transition schools are schools with roughly less than half of its graduating seniors enrollingin postsecondary education the following year.


helped choose, plus a financial aid package, makes the postsecondary option lastlonger and makes it more salient. The bridge to postsecondary becomes an easierone to take. The program also keeps students’ immediate costs down by offeringassistance during class and covering the application fees.

LifeAfterHighSchool was supported by the Ontario Ministry of Training,Colleges and Universities (MTCU) in partnership with the Ministry of Edu-cation (MOE). The program consisted of three hour-long workshops wherebygraduating seniors were guided through the college and financial aid applica-tion processes. The first workshop encouraged students to consider local post-secondary programs that they could get into based on their high school grades,as well as provided a simple financial aid calculator to demonstrate how they canafford to attend. The second workshop had students apply for real to collegesor universities, with the application fees covered from cutting and pasting theapplication number into the LifeAfterHighSchool website. The third workshophelped students open and get started on the Ontario Student Assistance Programapplication and send follow-up emails and letters to parents with instructions tocomplete the task. During these workshops, either external facilitators or teach-ers trained in the application processes were also available to help the students ifthey required it.

For students at low-transition schools that were randomly provided withLifeAfterHighSchool, postsecondary application rates increased from 64% to78%, while enrollment increased the following school year by 5 percentage points,with the greatest impact for students who were not taking any university-trackcourses in their last year of high school (a 9 percentage point increase inenrollment for them). The experiment provides a good example of collaborationbetween policy makers, academics and evaluators to produce evidence for whethera new program might be worth scaling up.

Our last example concerns the role that social support has in educationalachievement. It comes from a forthcoming paper, Oreopoulos and Petronije-vic (2016), which reports results from a field experiment involving first yeareconomics students at one of the University of Toronto’s satellite campuses inMississauga. In this study, students were randomized into a control group orone of two treatment groups in which students received additional support andencouragement. Students in the first treatment group received weekly email andtext messages throughout the academic year, designed to provide information,support, tips and encouragement in their academic endeavours. Students werealso able to respond to these messages and ask questions themselves. Studentsfrom the second treatment group received support from personal coaches—academically successful and keen upper-year undergraduate coaches who wereresponsible for meeting with their assigned students at least once per week andinstructed to engage with their coachees throughout the year in person, by text, byemail or via Skype to respond regularly to their academic struggles and questions.

Students from the first treatment group finished the school year with an averageeconomics’ grade 1.5 percentage points higher than students in the control group.


.2.4

.6.8

Pro

babi

lity

appl

ied

or e

nrol

led

Applied to collegeor university

Applied tocollege

Applied touniversity

Enrolled in collegeor university

Enrolled incollege

Enrolled inuniversity

Control mean [Adjusted] treatment mean

FIGURE 6 Estimated LifeAfterHighSchool program effectNOTES: These bar graphs report the results from the LifeAfterHighSchool project, as reported inOreopoulos and Ford (2016). The increase in probability of applying or enrolling from exposure tothe program is estimated from a probit model of the outcome variable regressed on a treatmentdummy along with fixed effects for cohort year and school. The sample comprises grade 12graduates (ND38, 130).SOURCE: Oreopoulos and Ford (2016)

And, students from the second treatment group finished the school year with anaverage 6 percentage points higher than students in the control group. Theseresults, while promising, highlight an interesting potential trade-off between theeffectiveness of interventions in social support and the cost/intensity of suchinterventions. The intervention involving a personal coach had a much greaterimpact on individuals’ academic performance, but involved a larger time cost onbehalf of the coaches. Coaches could help about only seven students with 10 hoursof work per week. In comparison, the intervention involving supportive text andemail messages had a much smaller effect on individuals’ academic performance,but had a marginal cost close to zero.

Employment:Another transition that behavioural economists consider closely is the transitionfrom unemployment to employment. As with the transition from high school topostsecondary, interventions incorporating insights from behavioural economicshave tended to focus on making the process simpler, more salient and easier tofollow. Currently, interventions to this effect are under way in Canada. We present


three of them by the Federal Ministry of Employment and Social Development(ESDC) and the Social Research and Demonstration Corporation (SRDC). Ad-ditionally, by presenting these interventions, we distinguish between what havebeen termed “low-touch” and “high-touch” nudges.

We first consider two pilot projects organized by ESDC’s Innovation Lab thatsought to increase the use of a new initiative known as Job Match (JM). JM isan employment service that both employers and job seekers can access throughESDC’s Job Bank (JB) database.21 The employer details the education, skills andexperience required for the job and JM uses an algorithm to match job seekersbased on these characteristics. However, in order for the JM service to work, jobseekers are required to input their skills, education and experience into the system(Parent and Audet 2016).

Enrolling in Job Match requires: (1) registering for an account on the JobBank website (2) clicking on the “Sign up” link and (3) uploading relevant skills,education and experience—a process that can take more than 20 minutes. Manyindividuals with JB accounts do not initiate the JM enrollment process, and ofthose who do, some do not provide all of the requested information and therebyfail to complete their account. ESDC cites behavioural barriers such as inat-tention to the JM option, inertia with having to take 20 minutes to enroll anddiscounting uncertain, long-run benefits from sign-up as possible explanationsfor the low take-up rates (Parent and Audet 2016). It is with the stated goal ofincreasing account completion rates to allow the JM service to facilitate morerapid job matches that the ESDC conducted the two pilot projects.

In the first pilot project we discuss, ESDC trialed five different designs thatdirected visitors from the JB database to the JM service webpage (figure 7) (Par-ent and Audet 2016). They were designed to compel more JB browsers to clickthrough to the JM webpage and consequently enroll in the JM service. Threeof the five designs included a different “nudge” statement, each drawing inspira-tion from one of three behavioural economic principles: developing social norms,priming/salience and framing.22 The designs were then trialed on a weekly rota-tion over four and a half months, allowing each of the five designs to be measuredfor a total of three to four weeks.23 To assess the relative success of each design,there were three outcomes of interest: (1) the rate at which visitors to the JBdatabase clicked through to the JM service webpage (2) the number of visitorswho started the JM account creation process and (3) the number of new JMaccounts completed.

Using Google Analytics, ESDC estimates that the most effective design—framing (design 4)—may have been responsible for 15% of new JM accountsduring the weeks it was trialed (Parent and Audet 2016). This was estimatedby measuring the increase in the proportion of JB visitors completing a JM

21 The Job Bank is a federally administered database that hosts searchable job listings fromemployers across Canada, delivered by ESDC.

22 For a detailed explanation of these behavioural economic principles, see Samson et al. (2015).23 A randomized control trial was not feasible given the nature of the JB and JM website

platforms, and thus, weekly rotations of the web designs were the next best alternative.


FIGURE 7 ESDC Job Match website link designsNOTES: ESDC trialed five different designs that directed online visitors from the ESDC Job Bankdatabase to the Job Match service page. Designs three, four and five include a different “nudge”statement, each drawing inspiration from one of three behavioural economic principles: developingsocial norms, framing and priming/salience.SOURCE: Parent and Audet (2016)

account when presented with the “framing” design, relative to visitors in thecontrol weeks. Similarly, the nudge statements that either primed the job seekerto consider the benefits of the JM service (design 5) or conveyed a social normsurrounding the use of the JM service (design 3) are also estimated to have hadcomparable effects. However, the “call to action” design, which focused only onthe salience of the design and did not include any accompanying nudge statement,


was less than half as effective as the other designs in encouraging job seekers tocreate a JM account (Parent and Audet 2016). The results of this pilot projectdemonstrate how small tweaks to modes of communication can substantiallyaffect individuals’ decisions. Moreover, because of this pilot project, ESDC hasmade the “framing” design permanent—a policy intervention that was not self-evident prior to the experiment.

To the same end, ESDC conducted another similar pilot project, this timesending emails to Canadians who had started their JM account but had not com-pleted it. These emails contained a link to the JM application page, encouragingrecipients to complete their JM account. Because the experimenters could cus-tomize the emails sent to individual recipients they were able to implement arandomized control trial (RCT) design, testing the effectiveness of different linksin compelling job seekers to complete their JM account. By monitoring the sentemails, the experimenters contrasted the rate at which email recipients clicked onthe different email links and the corresponding probability they completed theirJM account.

Throughout the study, four treatment emails and one control email weremonitored; among the treatment emails, four different behavioural economicprinciples were employed.24 Although the overall effectiveness of the emails wasmodest (a 5% increase in account completion), there was a profound differencein the click-through rates between those who received the control email and thosewho received the treatment emails.25 On average, the recipients receiving emailsincluding nudge statements were 77% more likely to click on the JM link thanthose who received the control email. Again, this pilot project demonstrates howsmall tweaks to an intervention can make a meaningful difference on the take-up rate of programs, and thus on individuals’ long-term well-being. Both pilotprojects by the ESDC show how policy makers can regularly test different de-signs and communication approaches based on behavioural economics to dis-cover what methods work best.

These interventions by the ESDC are a good example of what are termed “low-touch” nudges; the interventions involved small tweaks in modes of communi-cation for the job seeker. More generally, low-touch nudges are often cheaperto implement and focus on making decisions more salient and simpler for theindividual. This is in contrast to “high-touch” nudges, which involve significantchanges to one’s environment and often include personal interaction.

An example of a high-touch nudge is the application of motivational inter-viewing (MI). MI is a unique style of interviewing that is client-centred andexplorative; interviewees are encouraged to consider and develop resolutions to

24 The nudge statements: (1) framed the decision to complete the JM profile in terms of its benefitson the recipients’ employment prospects, (2) attempted to build a sense of commitment onbehalf of the recipients by highlighting how far they had already gone in setting up their JMaccount, (3) tried to inculcate a social norm by informing the recipients how many Canadianshad completed their JM account or (4) attempted to make the email link especially salient.

25 The click through rate is the percentage of email recipients who clicked on the emails links,directing them to the JM website.


problems themselves (Ford et al. 2014). One particularly distinct feature of MIis the assumption that interviewees each possess an intrinsic capacity to affectpositive change. MI then attempts to engender such change by encouraging in-terviewees to realize this capacity and to develop plans that achieve their desiredgoals (Ford et al. 2014).26 Here in Canada, MI has been trialed as a means toassist the transition to employment.

Working with funding from the Manitoba government, Canada’s Social Re-search and Demonstration Corporation (SRDC) piloted MI among a sampleof Employment and Income Assistance (EIA) recipients in Winnipeg, Mani-toba, from March 2015 to March 2016 (Palamar et al. 2016). In the pilot project,1,113 EIA recipients were assigned—at the EIA office level—to either a treatmentgroup in which participants partook in MI with their caseworkers or a controlgroup in which participants did not.27, 28 These treatment and control groups werecomprised of 13 Winnipeg EIA offices, in which each office’s caseworkers eitherreceived MI training or did not (Palamar et al. 2016). During the pilot project,EIA recipients received otherwise regular counselling from their caseworkers.Thus, under the assumptions that aside from partaking in MI, the clients in thetreatment offices were otherwise identical to those in the control offices and thatthere were no other factors common to each of the two groups of offices, the ex-perimental design allows for unbiased comparisons between EIA recipients whoreceived MI and EIA recipients who did not.

In this pilot project there were three key outcomes of interest: (1) EIA exit—theprobability participants continued to receive EIA benefits, (2) employment serviceusage—the probability participants started to receive employment services onemonth or more after the commencement of the pilot project and (3) pursuit ofgoals—the amount of progress participants made towards achieving goals whichthey specified prior to the initiation of the pilot project (Palamar et al. 2016).29

Most notably, Palamar et al. (2016) find that participants who were in officeswith caseworkers trained in MI were significantly more likely to exit EIA thanparticipants in offices with caseworkers who did not receive MI training; 9 to 12months after enrollment in the pilot project, participants in the treatment groupwere 6.8 percentage points more likely than participants in the control group to

26 For more information on motivational interviewing, see Miller and Rollnick (2002).27 Participants in the control group received intentional practice (IP) whereby their caseworkers

were encouraged to have deliberate conversations with the participants to help them reach theirgoals, but the caseworkers were not directed on how these conversations were to occur.

28 The caseworkers of five EIA offices received MI training and the caseworkers of six EIA did not.Further, historical statistics from the EIA offices were used to ensure similar clientcharacteristics across both the control and treatment offices. Palamar et al. (2016) also reportthat the baseline characteristics of the clients were very similar across the two sets of offices, andthey record adjusted versions of their estimates accounting for the slight differences in baselinecharacteristics between the treatment and control offices. Palamar et al. (2016) find that resultswere very similar across both the adjusted and unadjusted estimates and, hence, report only theunadjusted version of their estimates.

29 Prior to the pilot project, all participants were asked to identify specific steps they wanted totake, and these steps were categorized into nine separate goal types, ranging from language toeducational goals.


FIGURE 8 Effect of motivationals interviewing on the Employment and Income Assistanceprogram (EIA)NOTES: The data used in the presentation is from the Social Allowances Management InformationNetwork (SAMIN) research data set data for enrolled general assistance and single parent EIArecipients. Non-MI group (N D 533) and MI group (N D 399).SOURCE: Replication of the graph “EIA Exit Outcomes and Impacts – Trends,” found in slide 17of Palamar et al. (2016)

exit EIA (Palamar et al. 2016). Figure 8 shows how the difference in EIA exit ratesmanifested itself over time. Similarly, MI had a positive impact on participants’progress toward reaching their pre-specified goals.30 However, no identifiableimpact of MI on employment service usage was observed between the initiationof the pilot project and November 2015—eight months after the first EIA clientswere enrolled in the project.

To put these results in context, Palamar et al. (2016) estimate the net benefitof the pilot project to be $405 per program participant. This figure is drivenby reduced EIA payments to those in the treatment group relative to those inthe control group. Furthermore, this estimate was constructed using only thedifference in EIA payments over the pilot project period and does not account forthe fact that differences in EIA receipt likely continued after the end of the pilotproject. Palamar et al. (2016) thus conclude that MI has a significant positiveimpact on EIA and other indicators of goal attainment for many clients and

30 Palamar et al. (2016) report that participants in the treatment offices were 3.3 percentage pointsmore likely to make progress towards reaching their goals as specified by the participants priorto undergoing MI (Palamar et al. 2016).


further recommends that the training of MI be extended to all EIA client-facingstaff (Palamar et al. 2016).

Yet, the remarkable success of this project masks one important feature ofhigh-touch nudges such as MI: it is far more difficult to implement and evaluatehigh-touch nudges than low-touch nudges. High-touch nudges are necessarilymore involved than low-touch nudges, tend to be more expensive and are thusharder to trial on a wider scale, making it difficult to attain definitive results.This is perhaps best evinced by the fact that this particular pilot project actedas a follow-up to a previous intervention by SRDC that did not find conclusiveevidence that MI affected participants’ propensity to become re-employed.

Ford et al. (2014) report on this initial pilot project that involved a sample of155 long-term Income Assistance (IA) recipients in British Columbia. As with theaforementioned MI pilot project in Winnipeg, these long-term IA recipients wereassigned to either a control group that underwent regular employment counsellingor a treatment group that received employment counselling from caseworkerstrained in MI. Notably, Ford et al. (2014, p. 4) report that the program was “in-conclusive with respect to its impacts.” Here, the small sample size constrained thepower of the experiment, necessitating the follow-up study to clarify the potentialeffects of MI on re-employment rates. Since pilot projects involving high-touchnudges are often expensive and time consuming to administer there is potentiallylarge costs to trialing interventions large enough to effectively evaluate.

Taxation:Our last example of the application of behavioural economics to public policy isin the domain of taxation. This is an area that has received a lot of attention frombehavioural economists, signalling the potential for incorporating behaviouraleconomic insights into tax policy design. Researchers have shown that the salienceof tax rates, the ease of tax compliance and the framing of tax policy more broadly,all matter in determining our economic behaviour (e.g., Chetty et al. 2009, Chettyet al. 2013, Hallsworth et al. 2014).

Perhaps the most notable behavioural economic interventions in tax policyconcern tax compliance. Traditional policies to increase tax compliance typicallyinclude placing an economic cost—such as a fee—on non-compliance, but auditsand legal enforcement are often expensive to carry out. Yet, alternate policiesbased on findings from behavioural economics can achieve marked improvementson compliance rates at relatively low costs. For instance, in the UK, there havebeen field experiments showing that altering the wording on late-payment noticesemphasising high tax compliance among local residents are much more effectivein encouraging people to submit their taxes than regular late notices (BehaviouralInsights Team 2011). Further, simple and personalized text message remindersencouraging people to pay fines on time were effective in raising on-time paymentsduring another UK field experiment (Behavioural Insights Team 2011).

Inspired by the work of the UK’s BIT, the Canada Revenue Agency has alsoconducted similar experiments. In doing so, the Canada Revenue Agency has


(a)

(b)

FIGURE 9 Canada Revenue Agency’s signature block experimentNOTES: 9(a): Original signature block on Canadian personal income tax returns.9(b): Experimental signature block on Canadian personal income tax returns as implemented by theCanada Revenue Agency in a recent pilot project. This signature block was moved to the top of thetax return, its text enlarged, the wording surrounding the block simplified, the warning for providinginaccurate information bolded and tax filers were asked to print their name beside their signature.SOURCE: Authors’ correspondence with the Canada Revenue Agency

become a prominent exemplar for using RCTs in policy work. For instance, theCanada Revenue Agency has experimented with nudge statements in its lettersencouraging compliance with Tax Free Saving Account rules, tested the effec-tiveness of highlighting social norms in tax collection letters and trialed differ-ent messaging techniques encouraging the disclosure of unreported income byworkers in a subsector with higher-than-usual underground economic activity(Canada Revenue Agency 2014, 2015; Dutil and Mackey 2016).

A particularly interesting recent experiment by the Canada Revenue Agencyinvolves changing the signature block on the personal version of personal incometax returns, where tax filers attest to the accuracy of the information they submit.The experiment involves moving the signature block from the end of the form to itsbeginning. The text is also enlarged, the wording surrounding the block simplified,the warning for providing inaccurate information bolded and tax filers are askedto print their name besides their signature (figure 9). The rationalisation behindthis experiment proceeds from previous research suggesting people are more likelyto be truthful after they are prompted to think about honesty (e.g., Shu et al. 2012).

To evaluate the effectiveness of this intervention, paper tax returns containingthe altered signature block were distributed in two Canadian cities. The CanadaRevenue Agency is currently analysing the tax returns from these two cities,checking to see how they differ between the treatment group and a control groupcomprised of tax filers who filled out the regular tax returns with special atten-tion paid to parts of the return that are susceptible to dishonest disclosure byindividuals. Pending differences between the treatment and control groups, wemay see this intervention rolled out extensively in the future.

We think that this intervention is an especially nice one to finish with. It is aquintessential example of how incorporating findings from research inbehavioural economics into public policy can suggest policy changes that arevery cheap and have the potential to affect dramatic change. Even if alteringthe signature block compels a moderate increase in per person tax revenue, thiscostless intervention could garner substantial savings for the government.


4. Conclusion

Over the past decade there has been a tremendous rise in the application of be-havioural economics to public policy. This rise is a direct result of the purposefuldevelopment of governmental agencies dedicated to incorporating findings frombehavioural economics into public policy. Most formative in the proliferationof these governmental agencies was the establishment of the UK’s BehaviouralInsights Team (BIT). After witnessing successful and low cost public policy inter-ventions by the BIT, governments from around the world have sought to constructtheir own such agencies (Whitehead et al. 2014). That governments the world overhave looked upon the UK’s BIT for inspiration is evinced by the active consultingrole the BIT has played to many of these governments.31

The Economic and Social Research Council (ESRC) estimates that at least51 countries have initiated some type of centrally administered agency designedto promote the use of behavioural economics in public policy (Whitehead et al.2014). Aside from the BIT, the most notable federal level agency promoting theuse of behavioural economics in public policy is the US’s Social and BehavioralSciences Team (SBST). Established in 2014, and buttressed by a recent presi-dential executive order to encourage the use of behavioural economics in publicpolicy, the SBST seeks to integrate findings from behavioural economics intofederal government programs (SBST 2015).32 Since the first two years of its in-ception, the SBST has worked on a plethora of projects, ranging from promotingworkplace savings among Federal employees through an email campaign, to en-couraging truthful disclosure of self-reported sales of goods and services to thegovernment, by simply including a signature box on the top of an online data-entry form (SBST 2015).

Many subnational governments have also established similar agencies. Forexample, in 2012, Australia’s New South Wales (NSW) government started apartnership with the UK’s BIT that resulted in the establishment of its ownBehavioural Insights Unit (NSW Behavioural Insights Unit 2014). NSW’s Be-havioural Insights Unit runs its own trials and interventions as well as supportsother NSW government initiatives. Notably, one of the NSW’s Behavioural In-sight Unit’s stated goals is to “contribute to the global body of BI evidence”(NSW Behavioural Insights Unit 2014). This stated goal highlights an interest-ing feature typical to such governmental agencies: many of the trials and inter-ventions conducted by agencies such as NSW’s Behavioural Insights Unit are

31 The UK’s BIT has international offices in New York and Sydney and has collaborated with andconsulted for various federal, provincial and state governments (see, for example,behaviouralinsights.co.uk/who-we-work-with andbehaviouralinsights.co.uk/behavioural-insights-team-australia).

32 In September 2015, President Obama enacted an executive order encouraging the use ofbehavioural sciences to inform policy decisions across government departments (The WhiteHouse 2015).


publicly documented, promoting the dissemination and use of effective policyinterventions the world over.

There are also non-governmental initiatives and agencies dedicated to integrat-ing behavioural economics into public policy. At the transnational level, the mostnotable is perhaps the World Bank’s Global INsights Initiative (GINI). Launchedin 2015 the GINI consults for government clients and assists various World Bankteams in implementing and evaluating projects that use behavioural insights (TheWorld Bank 2016). Additionally, many international development organizationssuch as USAID, AusAID, UNICEF and the WHO have also started to applyfindings from behavioural economics into their practices (Whitehead et al. 2014).

To the same end, many private sector consulting groups and demonstrationcorporations have worked closely with government entities. For instance, here inCanada, BEworks—a Toronto-based behavioural economics consulting group—partnered with the Ontario Energy Board to investigate methods to promoteenergy conservation among its clients (BEworks 2014). Other organizations suchas Deloitte and the University of Toronto’s Behavioural Economics in Action atRotman research centre have also conducted work in similar contexts.

As is evident from the examples in the previous section, Canada is no excep-tion to the international trend in applying findings from behavioural economicsto public policy. At the federal level, the Privy Council Office established its ownInnovation Hub in 2015 with the goal of acting as a resource to other federaldepartments that seek to “adopt new and innovative approaches to solving com-plex policy, programming and service delivery challenges” (Privy Council Office2016). Among the Innovation Hub’s stated areas of expertise is the use of be-havioural economics. They presently have a number of projects completed andunder way, with published results expected in early 2017. In addition, the Inno-vation Hub has been involved in supporting the Public Health Agency of Canada(PHAC) on an interesting project focused on expanding the use of a mobile phoneapplication called “Carrot Rewards.” The Carrot Rewards application seeks topromote healthy living by offering users of the application loyalty points fromthe loyalty points provider of their choice if they undergo what is deemed tobe healthy and/or educational activities (British Columbia Ministry of Health2016). First implemented in March 2016 in British Columbia, and currently be-ing expanded by PHAC, the Carrots Reward mobile application is an exampleof how the government can potentially motivate healthy behaviour on a largescale.

As discussed in the previous section, the Canada Revenue Agency is enthusi-astically incorporating behavioural economics into its design of tax and benefitsadministration policy. While the Canada Revenue Agency has historically soughtto improve tax compliance through various initiatives, it has started through itsinnovation lab to implement more broadly pilot projects that make explicit useof findings from within the behavioural economics literature. Furthermore, thesepilot projects tend to be in the form of RCTs, allowing for rigorous compar-isons of alternative nudges. The pilot projects range from trialing automated


telephone messages reminding employers to submit their payroll and sales taxes,to the aforementioned signature block experiment. The work being conducted bythe Canada Revenue Agency is very exciting; since most citizens interact with theagency in one way or another, just tweaking certain aspects of these interactionshas the potential to garner tremendous savings for the government.

Another very recent development at the federal level is the establishment ofESDC’s Innovation Lab. Launched in April 2015, ESDC established its Innova-tion Lab, with a mandate to find innovative solutions to service delivery chal-lenges, while promoting greater integration between policy, program and servicedelivery. The Lab brings together key partners across ESDC and outside thedepartment, right from the outset, to develop new thinking on ESDC’s mostpressing problems and to foster sustainable solutions that are responsive to thechanging needs of Canadians. The Lab creates space and provides expertise infacilitation and design thinking to apply and test new methods, to prototype andquickly learn what works and what does not and to de-risk experimentation byputting both ideas and processes through iterative testing with clients.

Since its opening, ESDC’s Innovation Lab has focused on organizational de-velopment, scoping and designing projects, engaging employees and acting asa catalyst for innovation within the department. The Lab has benefitted fromexternal advice such as from the Mind Lab in Denmark as it establishes the fun-damentals of a design lab organization including governance, evaluation, com-munications engagement and project development. The Lab’s first official projectis to make the online application for the CPP the premier method for Canadiansapplying to the country’s pension plan.

In Ontario’s 2015 budget, the province officially announced the establishmentof its Behavioural Insights Unit, specifically citing the success of other countriessuch as the UK in implementing similar agencies (Ministry of Finance 2015). TheBehavioural Insights Unit operates within Ontario’s Treasury Board Secretariatand works with various ministries and agencies to develop services that take intoconsideration findings from behavioural economics. Despite the relatively quietannouncement proclaiming its establishment—capturing only half a page of theOntario Budget’s 426 pages—since its inauguration, the Behavioural InsightsUnit has published the results of three completed pilot projects. The projectsinclude attempts to increase the use of online renewals for licence plate stickers, toencourage Ontarian homeowners to conduct roofing inside the formal economy,and as already discussed, to encourage more organ donor registrations throughOntario’s prompted choice system. All of these projects are available on theirwebsite, complete with their experimental designs.

We believe it is important that Canada have its own departments and agen-cies dedicated to trialing new initiatives incorporating insights from behaviouraleconomics, even despite other jurisdictions the world over running similar initia-tives. This is because policy interventions incorporating behavioural economicsare very context specific; these interventions often occur within an existing policyframework, and it is not always obvious what the best way to alter the status quo


is. A good example of this is Ontario’s attempt to increase organ donation con-sent rates. Ontario had a very unique procedure to elicit organ donation consent,and there was no obvious way to simplify the process or to make consent appearmore attractive. A specific intervention was needed to find what process workedbest for Ontario.

Despite the growing enthusiasm for behavioural economics and its applicationto public policy, there is disagreement with regards to its promise. Critics arguethat its findings are often too trivial to substantiate meaningful positive change.They argue that relative to traditional economic interventions, accounting forbehavioural biases in policy formulation often has only a marginal impact onindividuals’ behaviour and may not be sufficient to affect systemic problems suchas poverty (e.g., Loewenstein and Ubel 2010). Under this lens, focusing on formu-lating and trialing different nudges may be distracting away from implementingtraditional and proven public policy interventions.

Here, our claim that behavioural economics is an add-on to neo-classicaleconomics, and not a substitute, is particularly important. We hence agree withBhargava and Lowenstein (2015) that behavioural economic models should notreplace traditional economic models in informing public policy but shouldcomplement them. In particular, we believe that the policy problem should informthe policy response; if the economic incentives are in place to encourage desir-able behaviour, but individuals are still not acting in their best interests because ofknown cognitive/behavioural biases, then perhaps it is necessary to inform futurepolicy interventions with behavioural economics. However, we are still in a learn-ing phase and more work is needed to understand how precisely to incorporatebehavioural economics in public policy.

References

ACSFA, Advisory Committee on Student Financial Assistance (2005) The Student AidGauntlet: Making Access to College Simple and Certain. Final Report of the SpecialStudy of Simplification of Need Analysis and Application for Title IV Aid.Washington: Department of Education

Akerlof, G. A., and R. E. Kranton (2000) “Economics and dentity,” The QuarterlyJournal of Economics 115(3), 715–53

Behavioural Insights Team (2011) Behavioural Insights Team Annual Update 2010–11.London: Cabinet Office Behavioural Insights Team

Benartzi, S., and R. Thaler (2007) “Heuristics and biases in retirement savingsbehavior,” Journal of Economic Perspectives 21(3), 81–104

Beshears, J., J. J. Choi, D. I. Laibson, and B. C. Madrian (2009) “The importance ofdefault options for retirement savings outcomes: Evidence from the United States.”In J. Brown, J. Liebman, and D. A. Wise, eds., Social Security Policy in a ChangingEnvironment (pp. 167–95). Chicago: University of Chicago Press

Bettinger, E. P., B. T. Long, P. Oreopoulos, and L. Sanbonmatsu (2012) “The role ofapplication assistance and information in college decisions: Results from theH&R Block FAFSA experiment,” The Quarterly Journal of Economics 127(3),1205–42


BEworks (2014) Analyzing and Nudging Energy Conservation and Demand ShiftingThrough Time of Use Compliance. Report prepared for the Ontario Energy Board,125 pp.

Bhargava, S., and G. Loewenstein (2015) “Behavioral economics and public policy 102:Beyond nudging,” American Economic Review 105(5), 396–401

British Columbia Ministry of Health (2016) “Canada’s first wellness rewards programlaunches in B.C.” Retrieved August 16, 2016, from archive.news.gov.bc.ca/releases/news releases 2013-2017/2016HLTH0013-000326.htm

Canada Revenue Agency (2014) “Reducing participation in the underground economy:Canada Revenue Agency 2014–2015 to 2017–2018.” Retrieved August 16, 2016, fromcra-arc.gc.ca/gncy/lrt/strtgy-eng.htmls

(2015) “Report on plans and priorities 2015–16,” available atpublications.gc.ca/collections/collection 2015/arc-cra/Rv1-12-2015-eng.pdf

Canadian Institute for Health Information (2016) “More organs needed than availablein Canada.” Retrieved August 16, 2016, from www.cihi.ca/sites/default/files/document/corr 2016 publicsummary en.pdf

Canadian Pharmacists Association (2012) “Leading pharmacy through changemanagement: A toolkit for assessing and supporting practice change.” Secretariat forthe Blueprint for Pharmacy Coordinating Office

Carroll, G. D., J. J. Choi, D. Laibson, B. C. Madrian, and A. Metrick (2009) “Optimaldefaults and active decisions,” The Quarterly Journal of Economics 124(4) 1639–74

Chen, F., and R. Stevens (2016) “Applying lessons from behavioural economics toincrease flu vaccination rates.” Health Promotion International. DOI:doi.org/10.1093/heapro/daw031

Chetty, R., J. N. Friedman, and E. Saez (2013) “Using differences in knowledge acrossneighborhoods to uncover the impacts of the EITC on earnings,” AmericanEconomic Review 103(7), 2683–721

Chetty, R., A. Looney, and K. Kroft (2009) “Salience and taxation: Theory andevidence,” American Economic Review 99(4), 1145–77

Choi, J. J., D. Laibson, B. C. Madrian, and A. Metrick (2002) “Defined contributionpensions: Plan rules, participant choices, and the path of least resistance.” InJ. M. Porterba, ed., Tax Policy and the Economy, Volume 16, pp. 67–112. Cambridge:The MIT Press

Corace, K. M., J. A. Srigley, D. P. Hargadon, D. Yu, T. K. MacDonald, L. R. Fabrigar,and G. E. Garber (2016) “Using behaviour change frameworks to improvehealthcare worker influenza vaccination rates: A systematic review,” Vaccine 34(28),3235–42

Department of Finance Canada (2016) “Canada’s finance ministers agree to strengthenCanada pension plan.” Retrieved August 16, 2016, from fin.gc.ca/n16/16-081-eng.asp

Dutil, P., and J. Mackey (2016) “CRA’s nudging ways.” Retrieved August 16, 2016, fromcanadiangovernmentexecutive.ca/cras-nudging-ways

Dynarski, S., and J. Scott-Clayton (2006) “The cost of complexity in federal student aid:Lessons from optimal tax theory and behavioral economics,” National Tax Journal59, 319–56

Ford, R., J. Dixon, S-w. Hui, and I. Kwakye (2014) Motivational Interviewing PilotProject Final Report. Social Research and Demonstration Corporation

Frederick, S., G. Loewenstein, and T. O’Donoghue (2002) “Time discounting and timepreference: A critical review,” Journal of Economic Literature 40(2), 351–401

Gabaix, X. (2014) “A sparsity-based model of bounded rationality,” The QuarterlyJournal of Economics 129(4) 1661–710


Gionet, L. (2015) “Flu vaccination rates in Canada,” Health at a Glance series. StatisticsCanada, catalogue no. 82-624-X

Hallsworth, M., J. A. List, R. D. Metcalfe, and I. Vlaev (2014) “The behavioralist as taxcollector: Using natural field experiments to enhance tax compliance,” NBERworking paper no. 20007

Johnson, E. J., and D. Goldstein (2003) “Do defaults save lives?,” Science 302(5649),1338–39

Kahneman, D. (2011) Thinking, Fast and Slow. New York: Farrar, Straus and GirouxKenter, P. (2015, January 26) “Advisers, SmartSAVER Program aim to increase uptake

of RESP benefits,” Financial Post. Retrieved August 16, 2016, frombusiness.financialpost.com/resp/advisers-smartsaver-program-aim-to-increase-uptake-of-resp-benefits/wcm/9a3cfb49-b4bf-4e34-b8e1-0f7694a4d46d

Krajibich, I., and M. Dean (2015) “How can neuroscience inform economics?,” CurrentOpinion in Behavioral Sciences 5, 51–57

Laibson, D. (1997) “Golden eggs and hyperbolic discounting,” The Quarterly Journal ofEconomics 112(2), 443–77

Lavecchia, A., H. Liu, and P. Oreopoulos (2016) “Behavioral economics of education:Progress and possibilities.” In E. A. Hanushek, S. Machin, and L. Woessmann, eds.,Handbook of Economics and Education, Volume 5, Chapter 1, pp. 1–74.North-Holland

Loewenstein, G., and P. Ubel (2010, July 15) “Economics behaving badly.” RetrievedJune 20, 2016, from: The New York Times from nytimes.com/2010/07/1y/opinion/15loewenstein.html

Madrian, B. C. (2014) “Applying insights from behavioural economics to policy design,”Annual Review of Economics 6, 663–68

Madrian, B. C., and D. Shea (2001) “The power of suggestion: Intertia in 401(k)participation and savings behavior,” The Quarterly Journal of Economics 116(4),1149–87

McClure, S. M., K. M. Ericson, D. I. Laibson, G. Loewenstein, and J. Cohen (2007)“Time discounting for primary rewards,” The Journal of Neuroscience 27(21),5796–804

Messacar, D. (2015) “Do workplace pensions crowd out other retirement savings?Evidence from Canadian tax records,” Analytical Studies Branch Research PaperSeries, catalogue number 11F0019M, no. 371

Milkman, K. L., J. Beshears, J. J. Choi, D. Laibson, and B. C. Madrian (2011) “Usingimplementation intentions prompts to enhance influenza vaccination rates,”Proceedings of the National Academy of Sciences of the United States of America108(26), 10415–420

Miller, W. R., and S. Rollnick (2002) Motivational Interviewing: Preparing People forChange, 2nd Edition. New York: Guilford Press

Milligan, K. (2002) “Tax preferences for education savings,” C.D. Howe InstituteCommentary. No 174. 22 pp.

(2016, June 21) “Making sense of the CPP expansion.” Maclean’s. RetrievedAugust 16, 2016, from macleans.ca/economy/economicanalysis/making-sense-of-the-cpp-expansion

Ministry of Finance (2015) Building Ontario Up: Ontario Budget 2015, 426 pp., availableat fin.gov.on.ca/en/budget/ontariobudgets/2015/papers all.pdf.Government of Ontario

Moore, K. D., W. Robson, and A. Laurin (2010) Canada’s Looming RetirementChallenge: Will Future Retirees Be Able to Maintain Their Living Standards UponRetirement. Toronto: C.D. Howe Institute


Mullainathan, S., and E. Shafir (2013) Scarcity: Why having too little means so much.London: Allen Lane

National Audit Office (2015) Automatic Enrollment to Workplace Pensions. London,UK: Department for Work & Pensions

Nayar Consulting (2013) SmartSAVER: Final Evaluation ReportNew South Wales Behavioural Insights Unit (2014) Understanding People, Better

Outcomes: Behavioural Insights in NSW . Premier & Cabinet, Behavioural InsightsUnit

O’Donoghue, T., and M. Rabin (1999) “Doing it now or later,” The American EconomicReview 89(1), 103–24

(2001) “Risky behaviour among youths: Some issues from behaviouraleconomics.” In J. Gruber, ed., Risky Behaviour Among Youths: An EconomicAnalysis, pp. 29–68. Chicago: University of Chicago Press

Oreopoulos, P., and R. Ford (2016) “Keeping college applications open: A fieldexperiment to help all high school seniors through the college application process,”NBER working paper no. 22320

Oreopoulos, P., and U. Petronijevic (2016) “Texting students to help achieve their goals”(in progress)

Palamar, M., M. McKerrow, K. Myers, and B. Palameta (2016) “Motivationalinterviewing in the context of social assistance: Evaluating the effectiveness ofmotivational interviewing in reducing social assistance dependence.” PowerPointslides. Presentation was provided to the authors through email correspondence withthe Social Research and Demonstration Corporation

Palameta, B., C. Vincent, and J.-P. Voyer (2011) Challenges to Canada’s RetirementIncome System: Simplified Enrolment in Retirement Savings Plans. Ottawa: SocialResearch and Demonstration Corporation

Parent, A., and M. Audet (2016) “Job match nudge trials: Job postings,” BehaviouralEconomics and Service Innovation Research Unit, Strategic Policy Branch, inpartnership with Employment and Social Development Canada

Privy Council Office (2016) The Innovation Hub: Our ventures. Retrieved August 16,2016, from pco-bcp.gc.ca/index.asp?lang=eng&page=innovation&doc=2-eng.htm

Robitaille, N., N. Mazar, C. I. Tsai, and E. Hardy (2016) “Nudging to increase organdonor registrations in a prompted choice context,” working paper, BEAR, RotmanSchool of Management, University of Toronto

Samson, A., D. Ariely, P. Barden, G. Fehr, T. Gohmann, M. Jager, . . . H. Scott (2015)The Behavioral Economics Guide 2015, available at behavioraleconomics.com/BEGuide2015.pdf

Samuelson, P. A. (1937) “A note on measurement of utility,” The Review of EconomicStudies 4(2), 155–61

Service, O., M. Hallsworth, D. Halpern, F. Algate, R. Gallagher, S.Nguyen, . . .E. Kirkman (2014) EAST: Four simple ways to apply behavioural insights. London:The Behavioural Insights Team

Shimazono, Y. (2007) “The state of the international organ trade: A provisional picturebased on integration of available information,” Bulletin of the World HealthOrganization 85(12), 901–80. Retrieved August 16, 2016, from who.int/bulletin/volumes/85/12/06-039370/en

Shu, L. L., N. Mazar, F. Gino, D. Ariely, and M. H. Bazerman (2012) “Signing at thebeginning makes ethics salient and decreases dishonest self-reports in comparison tosigning at the end,” Proceedings of the National Academy of Sciences 109(38),15197–200


SmartSAVER (2014) “Canada learning bond take-up map for Canada 2014.” RetrievedSeptember 11, 2016, from community.smartsaver.org/seize-the-opportunity/#take-up-map

Social and Behavioral Sciences Team (2015) Annual Report, available atsbst.gov/download/2015%20SBST%20Annual%20Report.pdf.Washington: Executive Office of the President and National Science and TechnologyCouncil

Sunstein, C. R. (2014) “Nudging: A very short guide,” Journal of Consumer Policy 37(4),583–88

Thaler, R. (2015) Misbehaving: The Making of Behavioral Economics. New York:W. W. Norton & Company

Treasury Board Secretariat (2016) Behavioural Insights Pilot Project – Organ DonorRegistration. Retrieved August 16, 2016, from ontario.ca/page/behavioural-insights-pilot-project-organ-donor-registration

Trillium Gift of Life Network (2015) “Organ donation: Opt-in vs. opt-out.” RetrievedAugust 8, 2016, from giftoflife.on.ca/img/Opt-In-vs-Opt-Out-Infographic-Dec2015.png

(2016) “Registration stats.” Retrieved August 16, 2016, from beadonor.ca/scoreboard

Whitehead, M., R. Jones, R. Howell, R. Lilley, and J. Pykett (2014) Nudging All Over theWorld: Assessing the Global Impact of the Behavioural Science on Public Policy.Swindon, UK: Economic & Social Research Council

White House, The (2015) Executive Order – Using Behavioral Science Insights to BetterServe the American People, available at whitehouse.gov/the-press-office/2015/09/15/executive-order-using-behavioral-science-insights-better-serve-american

Wolfson, M. (2011) Projecting the Adequacy of Canadians’ Retirement Incomes: CurrentProspects and Possible Reform Options. Montreal: Institute for Research on PublicPolicy

Woodford, M. (2012) “Inattentive valuation and reference-dependent choice,” workingpaper, Columbia University

World Bank, The (2016) Global INsights Initiative, available at live.worldbank.org/global-insights-initiative

Documents

Applying behavioural economics to public policy in Canada · Canadian public policy. We ﬁrst contextualize government policies that have incorporated insights from behavioural economics